In short
“AI watch out era” for marketers—Claude and ChatGPT are embedding watermarks in outputs, making it harder to copy/paste AI-generated text/images without detection.
Guests
Daniel Murray (Marketing Millennials; marketing creator/podcaster) and Jay Schwetzelson (Do This Not That; subjectline.com; marketing podcaster).
Key claims
Claude adds watermarks to images and weaves text watermarks directly into the writing; copying/pasting Claude text may be flagged as AI-written. ChatGPT image posts can be detected on platforms like LinkedIn, though reworking in Canva or screenshotting/resizing may reduce detection.
Notable examples
powdered sugar on a black shirt in New Orleans; empty-nester move (Tulane to LA); “test it and see” advice; mention of possible future watermark-stripping tools.
Written by AI. May contain mistakes. Listen to the episode to check what was said.
Chapters
Tap a time to open that second in VOPersonal Updates and Life Changes
0:39 to 2:23
Hosts discuss personal life changes as Jay becomes an empty nester.
“I'm here with the traveling man, Jay Schwelzen.”
AI Tools and Marketing Changes
2:23 to 3:41
Discussion on the impact of AI tools like Claude and ChatGPT in marketing.
“So here's the topic of the day, which I don't think people are aware of.”
An Overview of AI Watermarking
3:41 to 4:50
The hosts explain the implications of AI watermarks on marketers and content creators.
“Well, I think the image thing is really important.”
Strategies to Avoid AI Detection
4:50 to 6:27
Tips on how to modify AI-generated content to avoid detection as AI.
“It is doing horrible when I just take an image from chat that I made and put it into LinkedIn.”
Final Thoughts and Travel Anecdotes
6:27 to 10:46
Closing thoughts on AI watermarking and humorous travel stories.
“We'll remove the watermark and then you could use it.”
Transcript
Automatic transcript. May contain errors.0:07Welcome to a new special series called The Bathroom Break. That extra 10 minutes, you either have to listen to marketing tips or use the bathroom or both. But I don't recommend both, but that's your choice. This collab is going to be super fun. We have Daniel Murray from the Marketing Millennials and me, Jay Schwetelson, from the Do This Not That podcast and subjectline.com. Each episode in the series, we are going to go over quick tips about different marketing topics. And if you want to be in the bathroom, fine. Just don't tell us about it. Thanks for checking it out. We are back with another bathroom break.
0:41I'm here with the traveling man, Jay Schwelzen. And he's going through a major life change right now. Both of his kids are out of the nest, flying high. and he's been in New Orleans. He's going to go to LA. But I have a question. First, how is I going? And also, are you eating the beignets when you're out there? Well, that's a good question. So yeah, I just moved my son into college. He goes to Tulane. I'm moving my daughter into college, LA next week. And so I'm going to be an empty nester. First off, it's very ridiculous. Every person I come across go, are you okay being an empty nester? I'm like, what am I supposed to say to that?
1:20no, I'm not okay. And I'm going to run away. I just say, yeah, it's going to be fine. But yeah, I'm in new Orleans and we did get the years and I got powdered sugar. I was wearing a black shirt and I was walking around and I got powdered sugar all over me. And I looked like, like I did splatter paint. And Ali was like, you're embarrassing, bro. Like what's wrong with you? I'm like, it was really good. You look like you went to a college party. Yeah. Oh, I couldn't be last. So we dropped off. My son got him all ready to go. I'm like, it was early. It was like five or five. I'm like, do you want to come out to dinner with us?
1:59We're flying out the next day. And he's like, absolutely not. He's like, I want to call my friends. I'm like, no, but we're here. He's like, that's a you problem. So anyway, so now, Daniel, you and I would just hang out that much more. Oh, yeah. We hang out so much right now. So that, that, I don't know. All right, let's jump into this nonsense. So here's the topic of the day, which I don't think people are aware of. There is a major change going on with how we all are using AI. I call it the AI watch out era because we'll get into both Claude and Chachapiti, but both Claude and Chachapiti are doing things that in their outputs now, they're making it very hard for us to remove the fact that we used AI to create stuff.
2:48That was the worst explanation. So Daniel, you want to talk a little bit about like what Claude is doing first? Yeah. So Claude's adding watermarks and I think it's because like the European Union like made them do this, but this is going to like make it harder for you as a marketer or you as an individual to take images, create them and post them because they're going to have watermarks and people are going to know it's AI. I've always said from the beginning, like, AI should, and this is actually kind of good for marketers, I would say, because if you're a graphic designer out there, if you're a creative person out there, like, it's actually going to help you because you can, like, now, like, your images are going to stand alone if you can create images for yourself and be able to create original things that Chad, GPT, and Claude aren't creating for you.
3:40So I actually think this is a good thing for marketers, but it's actually going to be harder if you are only using AI to create images. What's your what's your take on it? Well, I think the image thing is really important. But the other big thing that Anthropic Claude, that's who makes Claude's Anthropic, is they're putting a watermark in text itself. So let's say you go to Claude and you ask it for, you know, help me write blah, blah, blah. Okay. They are putting in, not, not even in the metadata. Okay. They're putting in this, they're weaving into the text and the way the text is written, this watermark that if you just copy and paste text from Claude and put it into something else, okay, it is going to be able to be viewed as AI written.
4:28So really what we need to be thinking about is Claude is really a first draft place right or after you come up your first draft it's like your first edit then you take it somewhere else then you edit it some more but the the age of copy and paste from claude specifically text or images is over and it's worldwide it did start with the eu ai transparency stuff the requirements but now they're doing it for worldwide they're doing it right now and then the other pieces it's leaning over to chat gpt and daniel i don't know if you've seen this i'm just curious like I've tested creating images in chat GP2 and posting it like on LinkedIn or whatever.
5:07It is doing horrible when I just take an image from chat that I made and put it into LinkedIn. Are you seeing the same thing, even regardless of being identified as AI? It's actually with all AI platforms that LinkedIn social platforms are good at reading it as I take it out and put it in like a Canva or something like that. because in the back in the metadata of like the image it's showing like it's AI. So if you take it out and put it in Canva, but I'm guessing they're doing something extra now because people have found ways to get around this. I'm also wondering because there are ways to write in Claude and Chajiri, but you're bringing in a lot of your first draft and helping AI edit it.
5:54I wonder, are they marking up those things too? Because if you're writing a bunch of things on your own and you're not a grammar person and you're saying chat GPT because you edit this for grammar and stuff like that, I wonder if they're editing it or it's only editing it if you're saying write me this based on research. I wonder how far they're going down the line with that. Yeah, it's interesting because I think we're minutes away from there being a zillion tools where you can go out and strip out the claw text watermark. You know what's going to happen? It's like, oh, drop it in here. We'll remove the watermark and then you could use it.
6:32I mean, that nonsense is going to. I mean, they've done that for, I mean, CapCut used to have like watermarks. They used to have watermarks for like images. Like if you make them on other platforms and there's so many things that. And then also there's a lot of people watermark their like memes or watermark the images and all of those programs to take that off and strip theirs. So I feel like there is going to be ways to strip this. But I think ChatGPG and Cloud are doing what they, their output is a watermark. So they've cleared the standard that needs to be and whatever happens after, it's not their fault, I'm guessing.
7:12Yeah, and I'll tell you what I've seen that does work right now, even though there's a lot of write-ups that say it doesn't work. So if you take any image created in AI, whether it's, you know, claw design, attractive material, whatever. And then you do put it in Canva, modify it, you know, save it, post it. I've seen that not get identified as AI and it works really well. Also screenshotting an image that you create in the AI and then resizing it a bit and posting it. Everything they tell you online says that that doesn't work. But for me, I found with our clients that that is actually working in terms of it not being identified as AI garbage.
7:48I also read, which I have yet to see any data that supports it, that if you use Canva and use the background remover, the BG remover, which we use all the time, that that is now also creating another one of these synth ID moments where it's being identified as AI. but again i've seen no data support that that's an issue so right now i think the move is take it out of the ai tools bring it to canva or screenshot it and then roll with it and you seem to be doing fine yeah i mean the also the key with this era we're in is that you have to be paying attention to these type of news because you could be doing the same thing that was working yesterday and it's going to change tomorrow because there's watermarks and there's all these things so test it and see what's going to work and but also don't be the one caught out creating images and having watermarks all of them and then people knowing you're the one doing it yeah all right before we wrap up again you've also been on the road traveling a lot how have your travels been going have you been eating a lot of airport food uh it's everything been smooth sailing for you um we we're running into this problem where i've told ari um she probably get mad at me that i'm saying this but i've been telling ari like could you go get some snacks before we go to on the plane and she's like coming back with like taking like 30 minutes and like getting actual like food like from places that she's ordering and i mean like just like get snacks so like that we could bring on the plane and then like the gates they were boarding by the time she gets back and this has happened twice that she's done this and like i mean whatever even we get snacks meaning like get like cheese it's with a plane in the water that's all i'm asking for but uh we've run into the problem where like we're we're missed out boarding twice oh no that's amazing yeah yeah twice on this traveling trip because of this uh getting food well i i went um ali ali uh stained herself and she's like can you see if there's like a stain remover in the airport somewhere so I was about to board so I'm running the airport to find the scene removed her pen thing which I found I got back and then she had already she went and she boarded without me I was like that's not cool like on the plane I go I found it she's like sorry I boarded I'm like what I mean it could have been better I mean she she got the spot for you yeah but I still think it's funny like if you're looking for her and like where are you No text, nothing.
10:25What happened? I mean, funny the plane took off. Yeah, I was looking for staying. I did find it. Well, everybody out there, watch out for the watermarks. New Daniel Review for the Marking Millennials. Or leave me a review on Do This, Not That, and we'll see you at the next one. Daniel, come on, man. I got to get back to work. Get out of there. All right. While he's still in there, this is Jay. Check out my podcast, Do This, Not That, for marketers. Each week we share really quick tips on stuff that can improve your marketing and hope you give it a try. Oh, here's Daniel. He's finally out. Back from my bathroom break.
11:01This is Daniel. Go follow the Mark and Millet podcast, but also tune into this series. It's once a week, the bathroom break. We talk about marking tips that we just spew out and it could be anything from email, subject line to any marking tips in the world. We'll talk about it. Just give us a shout on LinkedIn and tell us what you want to hear. Peace out. Later.
From the publisher
Did you notice? The copy-paste era just ended. Jay and Daniel break down the "AI watch out era": Claude and ChatGPT are now baking watermarks into the text and images they generate, so your AI content can get spotted the second it goes live.
Also: why this might actually be good news if you can make original work, why Claude just became a 1st-draft tool instead of a shortcut, and the Canva and screenshot moves Jay swears are still sailing past the detectors. (Plus beignet-related powdered sugar chaos and a cautionary tale about boarding a flight without your travel buddy.)If you're a Marketer who wants to keep using AI without getting dinged for it, this episode is for YOU.Follow Jay:LinkedIn: https://www.linkedin.com/in/schwedelson/Podcast: Do This, Not That
Follow Daniel:YouTube: https://www.youtube.com/@themarketingmillennials/featuredTwitter: https://www.twitter.com/Dmurr68LinkedIn: https://www.linkedin.com/in/daniel-murray-marketing
Sign up for The Marketing Millennials newsletter: https://themarketingmillennials.com/
Daniel is a Workweek friend, working to produce amazing podcasts. To find out more, visit: https://workweek.com/




